Fish populations, gill net catches and gill net selectivity in the Zambezi and Chobe Rivers, Namibia, from 1997 to 2000

(1)

(2)

Fish populations, gill net catches and gill net selectivity in the Zambezi and Chobe Rivers, Namibia, from 1997 to 2000

Clinton J. Hay*, Tor F. Næsje**, Servatius Kapirika, Johan Koekemoer, Rita Strand, Eva B. Thorstad and Karstein Hårsaker**

* Directorate Resources Management

Ministry of Fisheries and Marine Resources, Namibia Private Bag 13 355 Windhoek, Namibia

**Norwegian Institute for Nature Research

Tungasletta 2, NO-7485 Trondheim, Norway

(3)

Norwegian Institute for Nature Research (NINA) issue the following publications:

NINA Project Report

This series presents the results of the institutes' projects when the results are to be made available in English. The series may include reports on original research, literature reviews, analysis of particular problems or subjects, etc. The number of copies printed will depend on demand.

In addition to this report series published in English, NINA publish the following series in Norwegian:

NINA Fagrapport (Scientific Reports)

This series present the results of NINAs own research work, overviews of problems, existing knowledge within a topic, literature reviews and material gathered from outside sources.

The reports are issued as an alternative or a supplement to international publication when timing, the nature of the material or the sector targeted, call for it.

NINA Oppdragsmelding (Assignment Report)

This is the minimum report supplied by NINA when exter-nal research or investigative assignments have been com-pleted.

The numbers of copies are limited.

NINA Temahefte (Topic)

These series present special topics. They are targeted at the general public or specific groups in the community, e.g. the agricultural sector, environmental departments, tourism and outdoor recreation, organizations etc. It is therefore more popular in its presentation form, with more illustrations than the previously mentioned series.

Most of the publications mentioned above are in Norwegi-an, but will also have an English abstracts and legends of figures and tables in English.

NINA NIKU Fact Sheet (Fakta-ark)

These double-pages sheets give (in Norwegian) a popular summary of other publications to make the results of NINAs and NIKUs work available to the general public (the media, societies and organizations, various levels of nature management, politi- cians and interested individuals).

In addition, NINA's and NIKU's staff publish their research results in international scientific journals, symposia proceedings, popular science journals, books, newspapers, and other relevant publications. NINA-NIKU also has a WWW home page:

http://www.nina.no

Hay, C. J., Næsje, T. F., Kapirika, S., Koekemoer, J. H., Strand, R., Thorstad, E. B. & Hårsaker, K. 2002. Fish populations, gill net catches and gill net selectivity in the Zambezi and Chobe Rivers, Namibia, from 1997 to 2000. – NINA Project Report 17: 1-88.

Trondheim, August 2002 ISSN: 0807-3082 ISBN: 82-426-1290 Management areas:

NINA•NIKU Foundation for Nature Research and Cultural Heritage Research

The report can be quoted with references to the source.

Editor(s):

Tor F. Næsje

Norwegian Institute for Nature Research Design and layout:

Kari Sivertsen

NINA•NIKUs Graphic Office Cover photo:

Tor F. Næsje Prints: Norservice A/S Stock: 150

Printed on environment paper Address to contact:

NINA•NIKU Tungasletta 2 NO-7485 Trondheim NORWAY

Tel: +47 73 80 14 00 Fax +47 73 80 14 01 http://www.nina.no

Accessibility: Open Project no.: 13530

Signature of responsible person:

Contracting sponsor:

Norwegian Agency for Development Cooperation (NORAD) Ministry of Fisheries and Marine Resources (MFMR) Norwegian Institute for Nature Research (NINA)

(4)

Preface

The White Paper “Responsible Management of the Inland Fisheries of Namibia” was finalised in December 1995, and forms the basis for a new law and regulations concerning fish resources management in the different freshwater systems in Namibia. Since all perennial rivers in Namibia border on neighbouring countries, management of the fish resources also depends on regional co-operation. It must also be taken into consideration that the fish resources might be exploited through subsistence, commercial and recreational fisheries. When implementing fisheries regulations for such complex systems, information on the fish resources and their exploitation are needed.

Based on a series of studies, recommendations will be given for management actions in the Zambezi and Chobe Rivers to involve local, national and international authorities and stakeholders, and to secure a sustainable utilisation of the fish resources for the benefit of local communities and future generations. The studies involve a description of parts of the recreational fisheries (Næsje et al. 2001) and of the migration and habitat utilisation of important fish species (Økland et al. 2000 and Hay et al. in prep). Økland et al. (2000) and an ongoing study (Hay et al. in prep) have shown that important fish species may perform migrations between countries. Furhermore, the biological and socio- logical aspects of the subsistence and semi commercial fisheries will be studied in 2001/2002. In the present report, the fish populations in the Zambezi and Chobe Rivers are described on the basis of six surveys performed in the period 1997 - 2000.

The project is a collaboration between the Freshwater Fish Institute of the Ministry of Fisheries and Marine Resources, Namibia, and the Norwegian Institute for Nature Research

(NINA). The work has received financial support from the Norwegian Agency for Development Cooperation (NORAD), the Ministry of Fisheries and Marine Resources and the Norwegian Institute for Nature Research.

We would like to express our gratitude towards the Director, Resource Management, Dr. B. Oelofsen and the Deputy Director, Resource Management, Dr. H. Hamukuaya for their support and encouragement during the project.

We are also thankful to Prof. Dr. P. Skelton and Mr. R. Bills from South African Institute for Aquatic Biodiversity (SAIAB, formerly: J.L.B Smith Institute of Ichthyology), Grahamstown, South Africa, who verified the identification of some of the fish species.

The Department of Water Affairs, Windhoek, provided the water level data of the Zambezi River.

The following staff members from the Freshwater Fish Institute were all involved in the field surveys or data punching:

T.P. Windstaan, Ms. S. Stein, J.H. Engelbrecht, J. May, A.

Mulundu, A.N. Mulundu, S. Beukes, the late S. Pootinu, S.

Jonas, A. Kahuika, J. Kahuika, E. Kahuika, the late F. Fillipus, N.

Lukas, E. Shikambe, E. Hayango, B. May and S. Hay. They are all gratefully acknowledged.

Windhoek/Trondheim, August 2002

C. J. Hay T. F. Næsje

Project leader, MFMR Project leader, NINA

(5)

1 Summary

Hay, C.J., Næsje, T.F., Kapirika, S. Koekemoer, J.H., Strand, R., Thorstad, E.B. & Hårsaker, K. 2002. Fish populations, gill net catches and gill net selectivity in the Zambezi and Chobe Rivers, Namibia, from 1997 to 2000.

NINA Project Report 017: 1-88.

Objective

The objective of this report is to provide baseline information about the fish resources in the Zambezi and Chobe Rivers to form the biological foundation for recommendations for a sustainable management of the fisheries. Based on fish survey data from the period 1997-2000, the fish resources are described through studies of species diversity, relative importance of the different species, life history parameters, catch per unit effort and selectivity of gill nets.

Methods

Fish were collected in five areas (Katima Mulilo, Kalimbeza, Lake Lisikili, Impalila and Kabula Bula) with survey gill nets (multi-filament, 22–150 mm stretched mesh size) and ten other sampling methods, such as seine nets, cast nets, electrofishing apparatus and rotenone. These are collec- tively called ”other gears” in this report. The gill nets were used to survey open, deep water habitats (> 1 m) in the main stream near the shore and deep backwater areas with some aquatic vegetation. The other gears targeted mainly small species and juveniles of long-lived species in shallow, vegetated and rocky habitats. Nordic multi mesh sized mono-filament gill nets were included during the survey in 2000 to improve sampling in the deep-water habitats and of the smaller species. Furthermore, fish caught during a fishing competition were sampled in 2000 to include biological data from larger specimens. Data from sampling with Nordic nets and the fishing competition were only included in the life history analyses of selected species, and not in other analyses. The restrictive use of these data ensures comparable data sets with previously reported Okavango River surveys, where these methods were not used (Hay et al. 2000)

Surveys were carried out three years in the spring during 1997-2000 and three years in the autumn during 1997- 1999. A total of 66875 fish were sampled, 39852 in gill nets, 7005 in Nordic nets, 562 during the fishing competition and 19456 with other gears. The most important species in the survey catches were identified by using an index of relative importance (IRI), which is a measure of the relative abundance or commonness of the species based on number and weight of individuals in catches, as well as their frequency of occurrence. Seventeen of the most important species collected were selected for a more detailed analysis of life history and gill net selectivity.

Results

A total of 69 fish species were recorded during the surveys, in addition to unidentified Synodontisspecies. Due to difficulties with the taxonomic classification in the Synodontis spp. group, these species have been pooled, except the easily recognised Synodontis nigromaculatus.

Seven Synodontis species have previously been listed for the Zambezi River, thus there may be up to six Synodontis species in the pooled Synodontisspp. group. The fish fam- ilies represented with the highest number of species were the Cyprinidae and the Cichlidae, with 20 and 17 species, respectively.

Six species were considered to be habitat specialists, which means their life history activities are confined to specific habitats, and that they required particular effort and equipment for collection. The habitat specialists recorded were Barbus codringtonii, Nannocharax macropterus, Leptoglanis cf dorae, Clariallabes platyprosopos, Chiloglanis fasciatus and Chiloglanis neumanni. Four of the species were difficult to find, whereas C. platyprosopos was common in its habitat.

Low numbers of Barbus kerstenii and Clarias stappersii were caught. These species are, therefore, also considered rare in the Namibian part of the Zambezi/Chobe Rivers.

Fourty species were caught in the gill nets (excluding Synodontis spp.). The ten most important species constituted 96 % of the total IRI. The two most important species (Brycinus lateralis and Schilbe intermedius) con- tributed to 73 % of the total IRI. The Characidae was the most important family in the gill net catches according to IRI (56 %), whereas the Cichlidae family constituted only a small part (2 %).

Sixty-seven species were caught with the other gears (excluding Synodontis spp.). The ten most important species constituted 74 % of the total IRI. The two most important species (Tilapia sparmanni and Pharyngochromis acuticeps) contributed to 30 % of the total IRI. In contrast to the gill net catches, the Cichlidae was the most important family in catches with other gears, according to IRI (58

%). The species diversity was higher for the catches with other gears than with gill nets, which is attributed to the flexibility of the other gears, and that a much wider range of habitats was sampled.

Thirty-six species were caught with gill nets at Kalimbeza, 33 species at both Kabula-Bula and Lake Lisikili, 28 species at Katima Mulilo and 24 species at Impalila (excluding Synodontisspp). Generally, ranking of the ten most important species in the gill net catches were corresponding at the different stations. When listing the ten most important species according to IRI at the five stations, only 15 species

(7)

were represented in total. According to IRI, B. lateralis and S. intermedius dominated the gill net catches at all stations, with the exception of the Lake Lisikili, where Petrocephalus catostomacontributed more in number and weight than S. intermedius. Species diversity in the gill net catches measured as the Shannon diversity index differed among stations, with the highest diversity in the Lake Lisikili and the lowest at Katima Mulilo. The year round presence of vegetation and lenthic conditions may have contributed to the high species diversity in the Lake Lisikili.

All the other stations included main stream habitats that usually yielded a lower catch and less variability in species.

Hydrocynus vittatus was absent at Kabula-Bula in the Chobe River, both in gill net catches and in catches with other gears. The backwater habitat at Kabula-Bula is considered less favourable for H. vittatus.

Among the ten most important species according to IRI, nine species were on the list both during high and low water. B. lateralis dominated the gill net catches during both periods. Water level had little effect on the species diversity in the gill net catches. However, three species had a marked decrease in the IRI from the high to the low water period, whereas six species had a marked increase.

The body length of the fish caught was up to 92 cm. The modal length of fish caught in gill nets was 8.0-8.9 cm, whereas for fish caught with the other gears 3.0-3.9 cm.

Thus, larger fish were caught with gill nets than with other gears, and this was true both for the species combined and for individual species. Twenty of the species caught had a maximum body length of 6 cm or smaller.

Of the selected species, twelve species had a minimum length at maturity smaller than 10 cm, two species between 11 and 20 cm and two species larger than 20 cm.

The minimum length at maturity was larger than or similar to the smallest fish caught with gill nets in all the selected species, except for both sexes of M. acutidensand males of P. catostoma. The length at 50 % maturity was larger than the minimum length of fish caught with gill nets for all the species of which 50 % maturity could be determined.

The 17 species selected for a more detailed data analysis, contributed to 93 % of the biomass of fish caught with gill nets and 56 % of the biomass of fish caught with other gears (one of the selected species was never caught in gill nets). These species represented a large variation in biology, distribution and sizes. Measured as numbers of fish caught per setting, the smaller gill net mesh sizes were the most effective in catching these species. For nine of the species, catch per unit effort in numbers was highest for the 22 or 28 mm mesh size, and for three of the species the 35 mm mesh size. Only two species were most effectively caught in the larger mesh sizes (57 and 73 mm).

Measured as weight per setting, larger mesh sizes were

more effective; six species were most effectively caught in the 22-28 mm mesh size, five species in the 35-45 mm mesh size and five species in the 57-150 mm mesh size.

For all species combined, the 28 mm mesh size was the most effective measured both as numbers of fish caught and weight per setting.

The Lake Lisikili station showed the highest catch per unit effort, both in terms of number of fish caught and weight.

The lake resembles a large backwater habitat, especially during flood, which may increase the productivity of the area. The lowest catch per unit effort in both number and weight was at Katima Mulilo, where the main stream habitat dominates. Main stream habitats are usually less productive than backwater and floodplain habitats.

The results did not show an unambiguous relationship between the catch per unit effort, habitat (mainstream ver- sus backwater) and water level (low water versus high water). Statistical analyses were carried out in all cases where comparable data for all mesh sizes existed, separat- ing the effects of station, habitat and water level.

Furthermore, comparisons were made for small mesh sizes (22 to 35 mm) and large mesh sizes (45 to 73 mm) separately, and for catch per unit effort measured in numbers and weight separately. Backwaters had in all cases a signif- icantly higher catch per unit effort than the mainstream - or no differences between backwaters and the mainstream were found. Regarding high and low water, no particular pattern could be seen.

Conclusions

The results from the surveys in the Zambezi/Chobe Rivers were compared with previous studies in the Okavango River (Hay et al. 2000). Generally, the fish fauna in the Zambezi/Chobe and Okavango Rivers showed great simi- larities, and there is a considerable overlap in the distribution of species between the rivers.

The complex and diverse nature of the fish fauna in the Namibian part of the Upper Zambezi has been revealed through the present surveys. However, detailed knowledge on the biology and behaviour of most of the species are still lacking. Basic information on life history, reproduction, movements, habitat preferences and habitat utilisation of target species is needed to regulate the fishery among the different countries and exploitation methods, and to evalu- ate the possible benefits of nature reserves and sanctuar- ies. The Upper Zambezi is presently still relatively undis- turbed by human impacts. For that reason alone, this system should be better studied to provide a baseline for future manipulations.

(8)

2 Introduction

Namibia is an arid country and strongly depends on the availability of open waterbodies for human food consumption, industries, irrigation and farming activities. The interior of the country has several man-made reservoirs, mainly for human consumption, where the largest is Hardap Dam in the seasonal southern Fish River. People in the north have to turn to fountains, boreholes, oshanas (shallow intercon- nected channels and pans) and perennial rivers to obtain potable water for their households. In the Caprivi Region, the Zambezi and Chobe Rivers play a significant role in the daily activities of the local communities such as fishery, agri- culture, transport, harvesting of vegetation and activities related to tourism.

Floodplain rivers, such as the Zambezi and Chobe Rivers, are among the most endangered ecosystems, and their fauna are especially under threat of species extinction and population disturbance (Halls et al. 1999). Multi-species floodplains with multi-gear fisheries have complex interac- tions between the environment, the fish communities and the fishers. Approximately 100 years ago, only 6000 people inhabited the Caprivi area (Mendelsohn and Roberts 1997).

At that time, the resources available could sustain the communities, and the anthropological impacts on the environment were insignificant. Today, the human population has increased 18 fold. All natural resources related to the rivers have been impacted by human activities such as farming, deforestation, building of roads, harvesting of vegetation for building materials and fisheries.

Historically, fishing was an important part of the ritual and political power base in the traditional management in the Caprivi region, and also today fish occupy a central place in people’s daily life (Tvedten et al. 1994). A common saying goes: ”If you don’t fish, you are not a Caprivian”.

Households eat fish daily for most of the year, and fish is the most important protein source ranked over beef, game and poultry (Turpie et al. 1999). Seventy-five percent of the households (Turpie et al. 1999) are engaged in subsistence fishing, with a mean reported catch of 370 kg per year per household (Turpie et al. 1999). A perceived decrease in the fish catches has been reported by the fishermen since the mid 1970’s.

The importance of the Zambezi and Chobe Rivers for the local communities cannot be over-emphasised. The fishery in the Caprivi is important for several reasons (Purvis 2001 a, b). The fishery provides a crucial source of protein, employment and income for households in the region. The trade in the fish products is especially important to the poorest households, which have no other means of gener- ating an income. A further important aspect is the barter of fish products for other essential commodities (Purvis 2001 a, b).

The fish resources in the Zambezi and Chobe Rivers are limited. As the local population grows and fishing activities increase, conflicts arise between subsistence, commercial and recreational fisheries. In addition, all the perennial rivers in Namibia, border on neighbouring countries. Management regulations and control measures are different in countries sharing the same fish resources. This has, among other problems, resulted in conflicts between foreign and native fishermen.

The objective of this report is to produce baseline information about the fish resources in the Zambezi and Chobe Rivers to form the biological foundation for recommendations for a sustainable management of the fisheries. Fish were collected in five main areas with survey gill nets and ten other sampling methods during 1997-2000. Based on these monitoring data, the fish resources are described through studies of species diversity in different parts of the rivers, the relative importance of the different species, the life history of important species and the catch per unit effort and selectivity of gill nets.

The stated policy in the White Paper “Responsible Manage- ment of the Inland Fisheries of Namibia” (Ministry of Fisheries and Marine Resources 1995) and the draft bill on inland fisheries, aim to ensure a sustainable and optimal utilisation of the freshwater resources, and to favour utilisation by subsistence households over commercialisation. The Zambezi and Chobe Rivers are shared with the neighbouring countries Botswana, Zambia and Zimbabwe. The fish resources play an important role in all these countries and should be co-managed to ensure the effective control of the fish resources to the benefit of all countries and communities. This report should not only benefit future management of the fish resources in Namibia, but also trans- boundary management actions of the freshwater fish resources in this region.

(9)

variable interaction between the Zambezi River and the Kwando-Linyanti-Chobe River systems. Lake Liambezi (Figure 3.1) was dry in the 1940s, filled up around 1952, and dried up again in 1986. However, the lake has partly been filled in 2001. The presence and the size of the lake are largely dependent on periods with floods and drought (Windhoek Consulting Engineers 2000). Flows in the Kwando River, which is the main source of inflow to Lake Liambezi, followed the patterns in the Zambezi River. Until 1999/2000, no significant floods have been recorded in the lower Kwando River since 1982.

The Zambezi-Chobe River systems consist of inter-linked rivers, backwaters, oxbow lakes, swamps and floodplains.

Aquatic plants prevail in the open water, which often is fringed by stands of reeds and water adapted grasses (Barnard 1998). The floodplains support a diverse grassland flora characterised by grass, shrub and herb species. The seasonal inundated floodplains form productive wetlands, which account for much of the species richness found in open waters.

Fishery and overgrazing of floodplains in the Eastern Caprivi are possibly the activities with the highest impact on the environment and fish community (Allcorn 1999). Pollution in the area is negligible. Large-scale development and urbanisation is not yet noticeable and, therefore, the physi- cal characteristics and water quality of each river system does not change drastically from one area to another. Dams and weirs do not occur along any of the parts of the rivers that were surveyed.

3.2 The Zambezi River

The Zambezi River is the largest river in southern Africa, draining an area of approximately 1.2 million km² (Timberlake 1997). The river rises in south-eastern Angola and in northern Zambia, and flows generally in a south- eastern direction. The Zambezi River reaches Namibia a few kilometres north of Katima Mulilo, forming the border between Namibia and Zambia for a distance of approximately 120 km to Impalila Island. The Chobe-Zambezi junc- tion is at Impalila Island, bordering the Chobe National Park in Botswana. From Impalila Island, the Zambezi River forms the border between Zambia and Zimbabwe (Figure 3.1).

The Victoria Falls form the barrier between the Upper and Middle Zambezi.

The river consists of a deep, wide mainstream, with bends and deep pools. Small-vegetated islands, sandbanks, bays, backwaters and narrow side streams occur frequently.

There are larger slow flowing channels, such as the Kalimbeza and the Kasai, and isolated pools. The only rapids are at Katima Mulilo and the Mambova Falls at Impalila Island.

3 Study area

3.1 The Caprivi Region

The Caprivi Region in Namibia is situated about halfway between the equator and the southern tip of Africa, and midway between the Atlantic and the Indian Ocean (Figure 3.1). The region borders on Botswana in the south, Angola and Zambia in the north and Zimbabwe in the east. The Chobe River and the Kwando/Linyanti System border on Botswana.

Within Africa, Namibia’s climate is second in aridity, after Sahara (Barnard 1998). Rainfall is lower and more variable than in the eastern subcontinent, and becomes lower and more variable towards the west. The country’s average annual rainfall is less than 250 mm, and the mean annual evaporation may be as high as 3700 mm in some areas.

The rainfall may be characterised as tropical semi-humid in the northeast, like in the Caprivi, to hyper-arid in the west (Figure 3.1). The Caprivi has the highest rainfall in Namibia, although a low rainfall in a global perspective.

The average annual rainfall at Katima Mulilo at the Zambian border in the Caprivi is approximately 680 mm, but has varied between 262 mm and 1473 mm during the past fifty years. However, it is important to note that the rainfall in the catchment area of the Zambezi River in Angola and Zambia is much higher, and that the rainfall in the Caprivi region has little effect on the water discharge in the river.

Six different land types are identified in the Caprivi (Mendelsohn and Roberts 1997). The largest portion of the region consists of the Kalahari Woodlands (55 %). The Caprivi Region is very flat, varying from 1100 m in the west dropping gradually to 930 m in the east, and elevations rarely exceed 30 m above sea level (Mendelsohn and Roberts 1997). Due to the flat topography and the presence of perennial river systems, especially the eastern parts expe- rience large annual flooding during summer and early win- ter. Floodplains cover 19 % of the Caprivi. In times of exceptional flooding, the Kwando - Linyanti and Zambezi - Chobe River systems are inter-linked, and large parts of the eastern Caprivi become one large floodplain (Curtis et al.

1998). In such cases, more than 30 % of the area east of the Kwando River becomes floodplains. The Caprivi wetlands have the highest overall species richness of the Namibian wetland systems, and 82 fish species occur in the Namibian part of this water system (Curtis et al. 1998). The floodplain ecosystems are complex and variable. Most Namibian fish species (78 %) are floodplain dependent for larval or juvenile stages and perform migrations between the floodplains and the main river (Barnard 1998).

The flat topography of the area creates a complicated and

(10)

17¡ 18¡15' 26¡

26¡ 10 km

17¡24¡ 24¡18¡15'

KalimbezaKatima Mulilo Victoria Falls

ZIMBABWE

BOTSW ANA

ZAMBIA

ZAMBIA NAMIBIA

Impalila Kabula-Bula N

ZambeziRiver

Chobe River

Lake Lisikili Lake Liambezi

KENYA

ETHIOPIA

SUDAN

EGYPT NIGERMAURITANIAMALI NIGERIASOMALIA NAMIBIA

LIBYA CHAD SOUTH AFRICA

TANZANIA

ZAIRE ZAMBIA BOTSWANAZIMBABWE

ANGOLA

ALGERIA MADAGASCAR

MOZAMBIQUE

South Atlantic

Indian Ocean

Figure 3.1. The Zambezi and Chobe Rivers in the north-eastern Namibia and the five main sampling localities (hatched areas).

(11)

In the mainstream of the river, sandy bottom substrate dominates. Muddy bottom substrate is often found in isolated pools, bays, backwaters and on floodplains where sil- tation occurs. Side channels and smaller side streams usually have a sandy bottom substrate. Few rocky habitats occur, except some near Katima Mulilo and Impalila Island.

Pebbles are found in the Impalila Island area.

Water flow in the river has been measured since 1907. The periods 1907-1950 and 1982-1997 were dry, whereas the period 1951-1989 had a higher rainfall (Figure 3.2 and 3.3). The water discharge normally peaks in April, although this may vary between March and May (Figure 3.4). The annual flood at Katima Mulilo has since 1907/08 fluctuated greatly, with maximum discharges between 870 m³/s (1914/15) and 8440 m³/s (1957/58) (Figure 3.2). In the same period, the annual water volume passing Katima Mulilo varied between 8300 million m³ (1914/15) and 68911 million m³(1968/69) (Figure 3.3). During our study period in 1997-2000, the peak water discharge at Katima Mulilo varied from 2070 m³/s in 1996/97 to 4541 m³/s in 1997/98 (Figure 3.5). The mean flood level is 5 m above the minimum water discharge, but has varied between 2 and 8 m (Tvedten et al. 1994).

The stream velocity varies from stagnant (backwaters and pools) to fast flowing water or rapids varying with the water discharge in the river. The narrow side streams are usually shallow and have a slow to intermediate flow. Side streams occur frequently during low flood, winding through sandbanks and islands. The water is clear with little suspended particles.

Floodplains occur during high floods, as inundated fields and grasslands. When the floods recede, isolated pools and backwaters are formed. Due to the confluence of the Zambezi and Chobe Rivers, a swamp like floodplain is formed in the Impalila Island area. The vegetation in this area consists mainly of dense impenetrable reeds, where Scrirpus sp. dominates (J.H. Koekemoer, pers. obs.). This area gives the impression of a swamped, vast floating mass of reeds, with interconnecting canals.

The river has ample available cover in the form of overhanging marginal terrestrial vegetation (submerged during high floods), marginal aquatic vegetation, and inner aquatic vegetation. Marginal terrestrial vegetation in the Zambezi River area can be described as fringing vegetation or riverbank cover in the form of terrestrial grass, reeds, overhanging

1907/081909/101911/121913/141915/161917/181919/201921/221923/241925/261927/281929/301931/321933/341935/361937/381939/401941/421943/441945/461947/481949/501951/521953/541955/561957/581959/601961/621963/641965/661967/681969/701971/721973/741975/761977/781979/801981/821983/841985/861987/881989/901991/921993/941995/961997/981999/00

Annual Flow Volume (Mm3)

Year 70000

60000

50000

40000

30000

20000

10000

0

Figure 3.2. Annual flow volume (Mm³) for the Zambezi River at Katima Mulilo (from Windhoek Consulting Engineers 2000).

(12)

1907/081909/101911/121913/141915/161917/181919/201921/221923/241925/261927/281929/301931/321933/341935/361937/381939/401941/421943/441945/461947/481949/501951/521953/541955/561957/581959/601961/621963/641965/661967/681969/701971/721973/741975/761977/781979/801981/821983/841985/861987/881989/901991/921993/941995/961997/981999/00 8000

9000

7000

5000

3000

1000

0 2000 4000 6000

Annual Peak Discharge (m3/s)

Year

Figure 3.3. Peak discharge (m³/s) for the Zambezi River at Katima Mulilo (from Windhoek Consulting Engineers 2000).

Mean Monthly Discharge (m3/s)

Nov 3500

3000 2500 2000 1500 1000 500 0

Oct Dec Jan Feb Mar Apr May Jun Jul Aug Sep

Figure 3.4. Mean monthly discharge (m³/s) for the Zambezi River at Katima Mulilo (from Windhoek Con- sulting Engineers 2000).

(13)

trees and shrubs. Vegetation can be dense in places, mak- ing the riverbank impenetrable. In other areas, grass and terrestrial reeds grow on sandy riverbanks and substitute the dominant dense vegetation of trees and shrubs, which grow on more stable ground (clay banks). Fallen or dead submerged trees also occur frequently along the banks.

Inundated grassland is the dominant floodplain vegetation.

Submerged trees and shrubs also occur on the floodplains.

Marginal aquatic vegetation is usually confined to the side of the mainstream due to fast or strong currents and consists of submerged aquatic vegetation, reeds and grass.

Aquatic vegetation is marginal in the mainstream, as it is difficult for the plants to anchor in the strong current. Inner aquatic vegetation occurs mostly in side channels and backwaters where there is low water velocity and possible for the root systems to fasten to the bottom. The aquatic vegetation consists of water lilies, water grass, reeds, and submerged and floating vegetation. Floating plants may occur anywhere in the Zambezi River, as it is dependent on wind and water currents for its distribution, and is most com- monly found in areas where there are obstacles preventing its further distribution. An alien species, Salvinia molestais a common example of a floating plant in the Eastern Caprivi (J.H. Koekemoer pers. obs.).

Where dense vegetation is present, the riverbanks show little signs of erosion. However, infrequent areas of erosion do occur, especially where deforestation has taken place by local fishermen and lodges to attain easy access to the river.

Erosion also occurs in the mainstream along bends, usually during high flood. Overgrazing also causes erosion in certain areas. The gradient of the riverbank varies from steep to moderate and low (J.H. Koekemoer pers. obs.).

3.3 The Chobe River

The Chobe River is a complex system consisting of a mainstream or channel, floodplains, backwaters and side channels. The river gets wider and deeper the closer it gets to the confluence with the Zambezi River. In the southwest near Ngoma, the river is narrow and channel like, but in the northeast near Impalila it develops into a wide, deep, strong flowing river. The mainstream has a low flow gradient, and the water velocity is low in most areas. The water is mostly clear with little suspended materials. Deep side channels and backwaters occur, where aquatic vegetation thrives.

Shallow floodplains occur in the southwest, whereas the floodplains in northeast are more swamp-like and deeper with dense reeds.

The direction of the water flow in the Chobe River changes seasonally, provided “natural” conditions with water in the Lake Liambezi, floods in the lower Kwando River and large floods in the Zambezi River. During high floods in the Zambezi River (February to May), the water might be pushed back up the Chobe River, some times as far as to the Chobe Swamps and the Lake Liambezi. However, in periods when the Lake Liambezi has been dry and with small floods in the rivers, only the Zambezi River has influ- enced the water flow in the Chobe River, forming a “backwater” to the main river.

Several types of aquatic vegetation are present. The water velocity is slow in places, especially during low flood, and it is possible for aquatic vegetation to fasten and grow in the mainstream. Aquatic vegetation may cover the entire riverbed, especially in the southern areas. Reeds, lilies, water Water flow (m3/s)

4500 4000 3500 3000 2500 2000 1500 1000 500

Jan Feb Mar Apr May Jun Month

Jul Aug Sep Oct Nov Dec 1996 1997 1998 1999 2000

0

Figure 3.5. Water flow (m³/s) in the Zambezi River (Katima-Mulilo) during the period 1996 to 2000.

(14)

grass and water ferns are abundant. Marginal vegetation such as water grass, reeds and submerged terrestrial grass occur all along the river. The fringing vegetation is domi- nantly grass and small shrub. Few marginal large trees occur on the Namibian side of the river. Dead submerged trees occur infrequently.

On the floodplains towards Ngoma, the vegetation is flooded field and grassland with shrubs and few trees. Near Impalila, the vegetation on the floodplains is marsh or swamp-like with dense reeds and a variety of submerged and protruding aquatic plants.

The riverbanks show little sign of erosion, and are well cov- ered with vegetation. The bank gradient is moderate to low.

3.4 The Lake Lisikili

The perennial Lake Lisikili is formed as a part of the Zambezi River. During high floods, water flows from the river into the lake and the Kalimbeza area. During low flood, the lake becomes isolated. The lake is approximately 4 km long and 0.5-1.0 km wide.

Various aquatic vegetation types are present in the lake.

The marginal aquatic vegetation is mostly reeds, while the inner aquatic vegetation is mostly submerged aquatic vegetation. Floating plants such as lilies and water grass is also abundant. Fringing vegetation is dense in areas, with numerous large trees. Terrestrial plants such as grass, shrubs and trees are submerged during high flood. The inundated grass and shrub create a floodplain like habitat on the margin of the lake. Rubble and gravel habitats are available with some small rocks.

The bottom substrate is mostly muddy with patches of sand. The water is clear. The lake deepens with a low gradient, but is deep in some areas (approximately 2m).

4 Materials and methods

4.1 Surveys and stations

Six surveys were conducted in the Eastern Caprivi during the period 1997-2000, of which three were during autumn (post-flood) and three during spring (pre-flood) (Table 4.1).

The five stations sampled were (1) Katima Mulilo and (2) Kalimbeza in the Zambezi River, (3) Lake Lisikili partly linked to the Zambezi River during high floods, (4) Impalila at the confluence between the Chobe and Zambezi Rivers, and (5) Kabula-Bula in the Chobe River. Stations are named after the closest village or known area.

Stations were chosen with respect to their commonness and similarity to the rest of the river system and its habitat types. Logistical difficulties such as distance, flood levels, accessibility and safety were taken into account when stations were selected. Stations include areas where possible external influences such as fishery, pollution, overgrazing and urbanisation could affect the ecosystem.

Table 4.1.Survey periods and total catch in numbers for the fish surveys in the Zambezi and Chobe Rivers in the period 1997 to 2000.

Survey year Spring* Total Autumn* Total catch (no) catch (no)

1997 X 10296 X 5980

1998 X 13195 X 8609

1999 - X 15187

2000 X 13608 -

* Autumn 97: 15 May -1 June, spring 97: 13-29 September, autumn 98: 7-28 May, spring 98: 5-25 October, autumn 99:

28 May-16 June, spring 00: 8-15 September and 5-19 November.

Some habitats (localities) occur seasonally and could only be surveyed during high or low floods. Examples of habitats that only occur during low floods are rocky areas and the habitats associated with it, isolated pools next to the river and certain backwaters. Some floodplains and large backwaters only occur during high floods.

4.2 Sampling design and methods

All stations were sampled with gill nets and several other gears (Table 4.2). A large range of gears and methods were used to limit the effect of gear selectivity and to survey all habitat types.

The gill nets were brown multi-filament nets with stretch mesh sizes from 22 to 150 mm (Table 4.3). The nets were

(15)

set from approximately 18:00 hrs in the evening to 06:00 hrs the next morning. At each of the five stations, gill nets were set at the same locality whenever possible during each survey. However, the variable water level caused sites to change with time. At some of the localities, gill nets could not be set during low water periods. The gill nets were used to survey open, deep-water habitats in the main stream near the shore and deep backwater areas with some aquatic vegetation. Nets were set either in the middle of a water-body or near marginal vegetation.

The other gear types were used at or close to the gill net localities. These gears targeted mainly small species and juveniles of long-lived species in shallow, vegetated and rocky habitats. The top layer of sandy substrates was also surveyed for species inhabiting these habitats. Nordic monofilament gill nets (Appelberg 2000) were included in the 2000 survey to improve sampling in the deep-water habitats and of the smaller species such as cyprinids (Table 4.2). Data from this method are only included in the life his- tory studies of the important fish species, and not in the

Table 4.2.Number of fish caught by gill nets, Nordic multi-mesh gill nets and other gears and methods used during the surveys in Zambezi and Chobe Rivers in the period 1997-2000. Specifications of the other gears used and total catches (number of fish) at the sta- tions sampled are also given. 1 = 15 m seine net, 2 = rotenone, 3 = 30 m seine net, 4 = D-net, 5 = traps, 6 = 2 m cast net, 7 = elec- troshocker, 8 = angling, 9 = hand scoop net, 10 = local gill net and 11 = 5 m seine net.

Station Gill nets Nordic Other Fishing Other gears used Total catches

gill nets gears comp.

Katima 4362 478 6386 1, 2, 4, 5, 6, 7, 8, 9, 11 11226

Lisikili 10819 0 4923 1, 2, 4, 5, 6, 7, 11 15742

Kalimbeza 9517 4729 4254 1, 2, 3, 4, 5, 6, 8, 9, 10, 11 18500

Impalila 4928 1798 1851 2, 4, 5, 8, 11 8577

Kabula-Bula 10226 0 2042 1, 2, 3, 5, 6, 7, 8, 11 12268

Zambezi 562 8 562

Total 39852 7005 19456 562 66 875

Table 4.3. Twine and mesh depth (number of vertical meshes) for gill nets of each stretched mesh size used during the surveys in the Zambezi and Chobe Rivers in the period 1997 to 2000.

Mesh size (mm) Twine Mesh depth

22 210D/4 158.5

28 210D/4 124.5

35 210D/4 99.5

45 210D/4 74.5

57 210D/6 59.5

73 210D/6 49.5

93 210D/9 42.5

118 210D/9 29.5

150 210D/9 24.5

rest of the analysis. In addition, in 2000 we also sampled all fish caught in a fishing competition in Zambezi River (Table 4.2) (Næsje et al. 2001). By doing this, we got data from larger individuals of several species, which had not been caught during ordinary surveys. These data are also included in the analysis of the life history of important fish species. This restrictive use of data from Nordic gill nets and the fishing competition ensures comparable data sets with the Okavango River survey, where these methods were not used (Hay et al. 2000). The different gears used at each station depended on the type of habitats present at the station.

The following gears were used:

•A 15 m seine net with a depth of 1.5 m, made from 30 % black shade cloth. This was used to sample shallower habitats such as backwaters, bays and also in the main stream, usually with a sandy or muddy substrate. It was occasionally used within aquatic vegetation.

•Rotenone was mainly used to survey rocky habitats. This was also the method used to collect fish from aquatic vegetated habitats.

•A 30 m seine net with a depth of 1.5 m, made from green anchovy net with a stretched mesh of 12 mm. This net was operated in large open water bodies with very little water flow. The substrate was usually sandy.

•A dip-net (D-net) was used in vegetated habitats and also in sandy substrates. The top 5 cm of the sand was exca- vated using the D-net to survey for Leptoglanisspp.

• A 2 m cast (monofilament nylon twine) net with a 20 mm stretched mesh was used to collect fish from deep-water habitats in backwaters and within the main stream. The water was either slow or fast flowing.

•A pulsed electrofishing apparatus (2 amperes and 600 volts) was used to sample rocky and vegetated habitats.

•A hand scoop net was used to sample fish within floating aquatic plants.

•A 5 m seine net with a depth of 1.5 m, made from 80 % green shade cloth. This was used to sample areas along

(16)

The common names and family classification for all the species (Appendix 1) are based on Skelton (1993). Seven Synodontisspecies are listed for the Zambezi River (Hay et al. 1999), but only one species, S. nigromaculatus, is easily identified morphologically. The other six species, S. leopar- dinus, S. macrostoma, S. thamalakanensis, S. vanderwaali, S. woosnami, and S. macrostigmawere, therefore, grouped together and recorded as one species group. When excluding the Synodontisspp. group, a total of 69 species were recorded in the total catches (Appendix 1).

4.3 Data collection and analysis

4.3.1 Biological data

Fish smaller than 100 mm in length were measured to the closest mm, while fish larger than 100 mm were measured to the closest cm. Fork length was measured on fish with a forked caudal fin, while total length was measured on fish with a rounded caudal fin. Fish weight was measured in the field as wet weight. Fish caught in gill nets were weighed to the nearest gram. Fish caught with other gears smaller than 50 g were weighed to the nearest 0.1 g, while larger fish were weighed to the nearest 1 g. After measuring and weighing a large number of individuals (often 50 or more), the remaining fish were separated into species, counted, pooled and weighed.

Sexual maturity was classified on a scale from 1 to 4 where 1 is immature or not developed gonads, 2 maturing gonads, 3 mature gonads and 4 spent fish.

4.3.2 Selected species

A large number of species (69 excluding Synodontis spp.) were caught in this study in the Zambezi and Chobe Rivers, and 17 species were selected for a more detailed data analysis (Table 4.6.) The main criteria for selecting these species were a) their importance expressed by the index of relative importance (IRI) in survey catches in gill nets and other gears, and b) their importance expressed by the numeral importance in survey catches in gill nets and other gears. The selected species represent a large variety in habitat use, distribution, trophic status, body size and general ecology. These species contributed 92.5 % of the biomass of fish caught in survey gill nets and 55.7 % of the biomass in other gears. One of the selected species, Aplocheilichthys johnstoni, was not caught in gill nets due to its small size.

For results dependent on gill net catches, number of selected species, therefore, are 16.

the river edges. The substrate was predominantly sandy with occasional mud.

•Conical-shaped traps were made from wire with approximately 2 mm mesh size. They were placed near the shore in shallow, strong water currents and within aquatic vegetation.

•Nordic monofilament gill nets were used to sample deep- water habitats. These nets consisted of 12 mesh sizes with the following panels: 86, 39, 12.5, 20, 110, 16, 25, 48, 32, 10, 70 and 58 mm stretched mesh. Each mesh panel was 2.5 m with a depth of 1.5 m.

•Angling with a rod and reel was used to catch larger fish.

A total of 66875 fish were caught with different gears during the surveys in the Zambezi and Chobe Rivers between 1997 and 2000 (Table 4.2, appendix 2a). Of these, 39852 fish were caught in survey gill nets and 27023 with the other gears.

The length data (appendix 2b) were based on measurements of 36834 fish. These fish were distributed among stations according to table 4.4, and among different fishing gears according to table 4.5.

Table 4.4.Length measurements of fish caught on different stations during the surveys in the Zambezi and Chobe Rivers in 1997 to 2000.

Station Length measured Total catch Percent of total

N N catch

Katima 6081 11226 54.2

Lisikili 6314 15742 40.1

Kalimbeza 11762 18500 63.6

Impalila 5400 8577 63.0

Kabula 6753 12268 55.0

Fish Comp. 524 562 93.2

Total 36834 66875 55.1

Table 4.5. Length measurements of fish caught by different gears during the surveys in the Zambezi and Chobe Rivers in 1997 to 2000.

Gear Length measured Total catch Percent of total

N N catch

Gillnet 21477 39852 53.9

Other gears 10401 19456 53.5

Nordic 4432 7005 63.3

Angling 524 562 93.2

Total 36834 66875 55.1

(17)

Shannon index of diversity (H´)

The Shannon index of diversity (H´) (2) is a measure of the number of species weighted by their relative abundances (Begon et al. 1990), expressed as:

where p_iis the proportion of individuals found in the ith species. The Shannon index assumes that individuals are randomly sampled from an ‘indefinitely large’ population, and that all species are represented in the sample. The value of the Shannon diversity index is usually between 1.5 and 3.5. A high value indicates high species diversity.

Index of evenness (J´)

Shannon’s index takes into account the evenness of the abundances of species, but we wanted a separate addi- tional measure of evenness. We used the ratio of observed diversity to maximum diversity to calculate the index of evenness (J´) (3) (Begon et al. 1990).

J is constrained between 0 and 1.0 with 1.0 representing a situation in which all species are equally abundant. As with H´, this evenness measure assumes that all species in the area are accounted for in the sample.

4.3.4 Gill net selectivity

Gill nets are selective fishing gears. This means that a specific mesh size catches fish in a certain length interval and is often most effective within a narrow length group. In addition, gill nets may discriminate among species according to fish morphology, for example body form and the presence of spines. Gill nets are also restricted to certain habitats, which will also influence the selectivity of this gear.

The body length distribution of fish in the different gill net mesh sizes is the simplest way to express and compare the gill net selectivity of different mesh sizes. For management purposes it is also necessary to calculate the gill net selectivity curve, which is an expression of the probability of cap- turing a certain size group of fish in a specific gill net mesh size. An analysis of body length distribution in gears, body length of mature fish and gill net selectivity are given for the 17 selected species (selected species, see section 4.3.2).

The general statistical model for gill net selectivity and its application are described in Millar (1992) and Millar and Holst (1997). When the actual distribution of fish in the sampled area is unknown, as in this study, selectivity esti- mates are based on the assumption that all fish have the same probability of encountering the gear. This may not always be true, as small individuals within a species may have different behaviour compared with larger ones. This uncertainty cannot be quantified without independent

J´=H´/H

_max

, where h

_max

=ln H´´ (3) H´= ∑p

i

ln p

i, (2)

4.3.3 Species diversity

Species diversity is defined as both the variety and the relative abundance of species. To calculate the relative importance and diversity of the different species, an index of relative importance (IRI) was used, as well as a measure of the number species weighted by their relative abundance, expressed as the Shannon diversity index (H`). An index of evenness (J`), which is the ratio between observed diversity and maximum diversity, was also calculated. Information about the species diversity in the Zambezi and Chobe Rivers were based on pooled samples from the five main stations.

Index of relative importance (IRI)

An “index of relative importance”, IRI, (1) (Pinkas et al.

1971, Caddy and Sharp 1986, Kolding 1989, 1999) was used to find the most important species in terms of number, weight and frequency of occurrence in the catches from the different sampling localities. This index is a measure of relative abundance or commonness of the different species in the catch and is calculated as:

where j = 1–S, %N_i and %W_i is percentage number and weight of each species in the total catch, %F_iis percentage frequency of occurrence of each species in the total number of settings and S is the total number of species.

IRI= (%

ⁱ⁺

W

i)

F

i

(%

^j+

%W

j)

F

j

100

(1)

Table 4.6.List of the ten most important species according to an index of relative importance (IRI) and numbers (No) in either survey gill nets or other survey gears from 1997 to 2000 (See Appendix 4 and 5). The species are ranked in accordance with their importance, and 1 is the most important species.

Species Gill nets Other gears

IRI No IRI No

Brycinus lateralis 1 1 9 9

Schilbe intermedius 2 3

Hydrocynus vittatus 3 6 6 Marcusenius macrolepidotus 4 4

Petrocephalus catostoma 5 2 Hepsetus odoe 6 10 Clarias gariepinus 7

Barbus poechii 8 5 7

Pharyngochromis acuticeps 9 7 2 6

Tilapia sparrmanii 10 8 1 1

Micralestes acutidens 9 10 3

Tilapia rendalli 3 4

Oreochromis macrochir 4 5

Pseudocrenilabrus philander 5 7

Barbus paludinosus 8 2

Barbus unitaeniatus 8

Aplocheilichthys johnstoni 10

(18)

information on population structure. This information, however, is rarely available and hard to obtain in natural fish populations. A further assumption is that all mesh sizes have the same efficiency on their optimal length class (the so-called ‘modal length’). This may also be erroneous due to different behaviour of small and large individuals. Often, the fishing efficiency may increase with mesh size. Several statistical methods are developed to represent the selection curves. Two functions were used in this study. The standard normal function was applied for species that are mainly entangled by their gills, whereas a skewed normal function (Helser et al.1991, 1994) was used for species that to some extent can be caught in other body structures such as fin rays, teeth and spines. The selection curves were standard- ised to unit height by dividing the number of fish in the modal length class.

4.3.5 Catch per unit effort

When standard fishing gear is used, the catch per unit of effort may be used as a rough indicator of the density of fish in the sampled area. For a standard series of gill nets in this study, catch per unit effort (CPUE) was defined as the number or weight of fish caught in 12 hours of fishing in a panel size of 50 m²gill net.

Measuring catch in number or weight of fish may give very different results. In this report, the results are generally pre- sented in both units, but with an emphasis on weight, as this unit is more important to fishermen and managers.

4.3.6 Databases and software

All recorded data were compiled in PASGEAR (Kolding 1995), which is a customised data base package intended for experimental fishery data from passive gears. The package is primarily developed to facilitate the entering, storage and analysis of large amounts of experimental data. The program makes data input, manipulation and checking data records easy. PASGEAR also contains predefined ex- traction, condensing and calculation programmes to facilitate data exploration and analysis from survey fisheries.

PASGEAR (version May 2000) and SPSS for Windows (version 9.0) were used to perform the calculations and statistical analysis.

5 Background biology and distribution of selected species

As a background to the results and discussion in this study, an overview of the biology and distribution for the most important species found in the surveys is given in this chap- ter. The information is mainly collected from Skelton (1993). The reference under the separate species is given only when information is collected from other sources. The species are classified according to family.

Cichlidae

Pseudocrenilabrus philander (Southern mouthbrooder) is widespread in Southern Africa from the Orange River and northwards to Malawi and the southern tributaries of the Zaire River. Several isolated populations are present in Namibia, such as in the Lake Otjikoto and the Otavi foun- tain. It may reach a length of 13 cm and breeds from early spring to late summer. It is a mouthbrooder with several broods raised in one season. This species lives in a wide variety of habitats, but prefers vegetated areas, feeding on insects, shrimps and even small fish. It is an aquarium species, and is also used in behavioural and evolutionary research.

Pharyngochromis acuticeps (Zambezi happy) occurs in the Okavango and Zambezi Rivers, but is absent from the Kunene River. It may grow to 22 cm, but is usually less than 10 cm. It is a female mouthbrooder and breeds in the summer. It occurs in a wide range of habitats, but needs cover such as vegetation or tree roots. It preys on insects, shrimps, small fish, and eggs and larvae of nesting fishes. It is a potential aquarium species.

Tilapia sparrmanii (Banded tilapia) is widespread in Southern Africa, with a similar distribution as the Pseudo- crenilabrus philander, and it has been extensively translo- cated south of the Orange River in the Cape. Individuals have also been translocated to several waterbodies in Namibia (Hay et al. 1999). It attains a length of approximately 23 cm and weighs up to 0.5 kg. It is tolerant of a wide range of habitats, but prefers quiet or stagnant waters with vegetation, where it feeds on algae, soft plants, inver- tebrates and small fish. It is common in subsistence fisheries and occasionally in angling.

Tilapia rendalli(Redbreast tilapia) is widespread in southern Africa where it occurs in the Kunene, Okavango and Zambezi River systems, in the eastern Zaire basin and in coastal rivers south of the Zambezi. It is also translocated to many catchment areas in southern Africa. It has also been recorded from the Lower Orange River and several waterbodies in Namibia (Hay et al. 1999). This species grows to about 40 cm and 2 kg, and breeds and raises several broods

Fish populations, gill net catches and gill net selectivity in the Zambezi and Chobe Rivers, Namibia, from 1997 to 2000